[ad_1]
re: Invent Amazon has designed its own 64-bit Arm Server processors, dubbed Graviton, and currently rents them on AWS.
Just in time for its annual re: Invent conference in Las Vegas, Internet titan has unveiled today its family of EC2 A1 instances, ranging from a1.medium to 1 vCPU, 2 GB of RAM and up to 15 GB. at 3.5 Gbps EBS and 10 Gbps of network bandwidth, which costs $ 0.0255 per hour on demand, up to 1.4 × large with 16 vCPUs and 32 GB of RAM, at a price of 0.408 USD per hour.
This may be less expensive than, for example, comparable T instances in terms of CPU; however, the specifications differ between the types of instances. You can compare and contrast for yourself here. Amazon estimates that A1 instances are up to 45% cheaper than their x86 virtual machines, depending on the configuration.
We performed a basic comparative analysis of a1.4xlarge versus c5.4xlarge using the Phoronix C-Ray reference. Render took 108 on c5, 112 on a1. c5 is $ 0.68 / hour, a1 is $ 0.408 / hour. It looks like it's really a good deal. #reinvent
– rbranson: executive vice president, thoughts (@rbranson) November 27, 2018
In any case, you should not just look at the raw price: you must test the performance and compatibility of your software stack on the A1 virtual machines before you engage in the architecture, because of the differences between Arm and x86.
The Graviton processors are AArch64, running Amazon Linux 2, Red Hat Enterprise Linux 7.6 or Ubuntu and are available in the Amazon East (North Virginia and Ohio), Western US (Oregon) and Europe (Ireland) regions in various forms from on request at the dedicated accommodation. Think of scalable platforms, web hosting, etc.
If you use elements written in portable or interpreted scripting languages, you can probably simply transfer this code to A1 virtual machines and test it. Otherwise, spit out this Armv8-A tool chain, retrieve pre-built packages from your favorite Linux distribution, or try pre-built binaries, depending on availability.
"If you do not look closely, you may not notice at the beginning that you are using an Arm Server, it was intended," said Red Hat's Jon Masters, who worked to ensure that RHEL is supported by A1 instances. "I've been playing with Amazon's A1 instances for a while and I'm blown away by the quality of the engineering and the capabilities of the team."
What is the ..?
If you're wondering where it all came from, well, in 2014, we warned that Amazon was thinking about using Arm chips for cloud computing resources. A major step in this quest has been to engulf Annapurna Labs for a few hundred million dollars in 2015. That biz is an Arm-System on-chip designer and licensee, initially focused on projects related to the Internet of Things. objects. Then, AWS revealed that it uses custom chips designed by Annapurna and nicknamed Nitro in its cloud. These perform network and storage tasks unloaded by EC2 hypervisors, allowing x86-based instance hosting machines to focus on executing client workloads while Nitro ASICs forwarded packets and data.
Today, the chip design team used this experience to design and deploy Graviton Arm processors in the AWS cloud, so that they execute client code on multipurpose virtual machines. . As one could expect from the Amazon, which remains very secret, the technical details available are not more numerous. We caught AWS PR managers for more information.
In the meantime, looking / proc / cpuinfo on an instance A1, the ID number of the CPU is 0xD08, which suggests that the Graviton is based on the Cortex-A72 plans of 2015 from Arm. We also note that a vCPU is mapped to a physical kernel. The 16 instances of vCPU are fixed in four quad-core clusters with 2 MB of shared L2 cache per cluster, and 32 KB of L1 data cache and 48 KB of L1 instruction cache, per-core.
Last month, Arm announced that one million Arm-powered data center servers would be delivered by 2018, and most of them were converged systems, which allows us to predict that a million of data center servers powered by Arm would be delivered in 2018, and most of them were converged systems, which allows us to predict that Amazon would count for a good part. And Amazon also offers AMD Epyc processors in its EC2 cloud. Meanwhile, Microsoft Azure wants to be at least 50% powered by the arm.
This puts additional pressure on Intel. According to analysts, Chipzilla's x86 processors account for more than 90 percent of the global data center computer market. It seems that Cavium's Epyc, Graviton, Thunder X2 companies and other competitors are reducing their share of Intel's market here and there. Intentional word game. ®
[ad_2]
Source link
